Article

What Apify can access when you connect Notion

For teams sending Apify results to Notion, learn what the connector can access, how to restrict its tools and what recurring usage costs.

Editorial illustration for What Apify can access when you connect Notion: a document represents the research briefing. Not documentary evidence.

For teams sending collected data into Notion, Apify’s MCP connectors let a compatible Actor—an automation program—write results without receiving the Notion token. The credential stays with Apify’s proxy, but the automation still uses permitted tools within the connected user’s access. Costs depend on the Actor and its usage, so a working connection does not establish that the workflow is free.

Apify announced the connectors on June 9, 2026. They use Model Context Protocol (MCP), a standard for software to call external tools. The decision for a Notion workspace owner is whether the automation’s access and recurring bill fit the job.

What controls Notion access?

Apify’s proxy keeps the Notion credential out of the Actor, while permitted tools can still act on accessible content. According to Apify’s documentation, the Actor uses an Apify run token to reach that proxy. It can use only connectors supplied in its input, and the proxy session expires when the run ends.

The underlying content boundary comes from Notion. Its user-authorized connection guide says a connected client can read and update content the user can access in the selected workspace. Apify’s tool restrictions can narrow the operations available within that boundary.

The practical implication: selecting an output database is not the same as restricting all access to that database. For example, Notion’s fetch tool can retrieve an accessible page by its URL or ID. If that tool is allowed, removing search alone does not make other accessible pages unreadable.

Nor does hiding a token keep retrieved data inside Notion. Notion’s security guidance says connected clients may send returned content to outside systems and recommends reviewing actions and data sharing. Before authorizing a workflow, check both the account’s reachable content and what the Actor does with it.

Where do you restrict the tools?

Apify documents three permission layers: the access granted during authorization, the connector’s allowed-tool list and the Actor’s declared tool constraints. Every configured layer must permit a call. Crucially, the connector allows all discovered tools by default; an Actor can also declare broad tool access.

  1. In Apify, open Settings → API & Integrations → MCP connectors and add or edit the connection. For manual setup, Notion’s current server address is https://mcp.notion.com/mcp. Apify handles the application setup for Notion’s OAuth sign-in and consent flow; you do not need to register a separate app.

  2. After authorization, inspect the discovered tools and restrict the connector to the operations the workflow needs. Ask the Actor developer which tools are required rather than copying a generic list.

  3. Check the Actor’s own declared permissions before selecting the connector in its input. A connector restriction applies across Actors using that connection; an Actor’s narrower declaration adds another limit.

Keep completion checks in mind when narrowing permissions. Notion’s page-writing documentation describes operations that continue in the background. An Actor using those operations may need notion-get-async-task as well as its writing tools to confirm the result.

Ending a run does not delete the saved connection. To withdraw that connector from all Actors, delete it in Apify’s connector settings; Apify says removal takes effect immediately.

Will your chosen Actor work with Notion?

Connecting the account is only one prerequisite. The Actor must be built to accept MCP connectors and implement the Notion delivery you need. An existing scraper does not gain that logic simply because you authorize Notion. Check its input form and documentation for connector support, destination selection and how results map to Notion properties.

Also verify the required tools on the intended connection. Notion documents restrictions by permission, connection capability and workspace plan; its notion-get-tool-access tool can check availability. A universal minimum plan for all write workflows is not established by these sources. Check the specific operation and options before choosing a plan.

If you are comparing automation platforms, the Activepieces guide to external MCP sign-in support covers another compatibility check: whether the platform can complete the external service’s authorization flow.

What does the Apify-to-Notion workflow cost?

As checked on November 16, 2026, Apify’s public monthly-billing prices in US dollars are:

  • Free: $0 subscription, with $5 of usage each month.

  • Starter: $19 a month, including $19 of usage, with additional pay-as-you-go charges when usage exceeds that allowance.

Those are Apify account allowances, not Notion licences or a guaranteed number of imported records. Under Apify’s subscription rules, Free cannot continue once its allowance is exhausted until the next cycle. Paid accounts must have a sufficiently high usage limit to continue into overage. Unused prepaid usage expires at the end of the billing cycle.

Illustrative monthly arithmetic, not a measured workflow bill: assume all eligible Apify usage totals $12 on Starter. The subscription still costs $19. If usage totals $30 and the spending limit permits it, the bill is $19 plus $11 in excess usage: $30, not $49. These examples exclude taxes, add-ons and external services.

Next inspect the chosen Actor’s Pricing tab. Store Actors can charge for events such as producing results, or for platform resources. Most event-priced Actors include platform usage, but some charge it separately. Do not add compute twice when it is already bundled; later dataset reads or writes can still incur charges.

For a resource-priced workload, Free and Starter list $0.20 per compute unit—one gigabyte of allocated memory for one hour. That is only compute, not an all-in import price. Other applicable platform resources and any external service charges still need accounting for.

The published allowances alone cannot establish the full cost of a Notion-writing Actor, including any connector-specific charges. Confirm the charges for your setup. The Firecrawl MCP access-and-usage guide makes a useful adjacent comparison if you are choosing a web-data source: connection access and funded recurring usage are separate questions.

What should you check before scheduling imports?

Before scheduling, run a limited batch with an explicit spending limit. Inspect the resulting Notion records and the account’s charges, then estimate recurring consumption from your own workload.

Ask the developer how the workflow handles incomplete writes and repeat runs. Notion can return an asynchronous task or warnings; a submitted request is not proof that the intended records are complete. For background writes, the Actor should wait for a final success or failure and inspect the result. Decide whether subsequent imports should create new records or update existing ones, and require a way to avoid unwanted duplicates.

Volume matters too: every Notion MCP tool call counts toward API request limits. Notion automatically retries a rate limit only once when the required wait is two seconds or less; otherwise it returns a tool error. A scheduled Actor needs to handle that outcome rather than assuming every batch will finish in one pass.

Use the connector when the Actor supports your destination, its permitted operations and reachable content are acceptable, and the budget covers recurring use. If the requirement is access to exactly one database, do not approve the workflow on the strength of token isolation alone: establish the content restriction separately.